Add drawBitmapLattice() API

Add drawImageLattice() and drawBitmapLattice() APIs

The specified image/bitmap is divided into rects, which
can be draw stretched, shrunk, or at a fixed size.  Will be
used by Android to draw 9patch (which are acutally N-patch)
images.

BUG=skia:
GOLD_TRYBOT_URL= https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&issue=1992283002

Committed: https://skia.googlesource.com/skia/+/c573a40ed5024b463e47088d307e3164a486dba5

[msarett, 2016-05-19 17:18:59]

Description was changed from

==========
Draft API for drawBitmapNine()

BUG=skia:
==========

to

==========
Draft API for drawBitmapNine()

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[msarett, 2016-05-19 17:22:52]

msarett@google.com changed reviewers:
+ djsollen@google.com, reed@google.com

[msarett, 2016-05-19 17:22:52]

Let me know what you think.  Once we're happy with this, I'll apply similar
changes to drawImageNine() and actually write the implementations.

I've left out "color hints" initially, since these are actually a performance
optimization and not necessary to draw correctly.  We can experiment with adding
these in once we have working code and performance benchmarks.

[reed1, 2016-05-19 17:39:30]

1. I suggest we keep the old version as well (even w/ the same name)
2. Lets try to define an object that holds all those arrays/counts/etc.

[msarett, 2016-05-19 18:01:27]

On 2016/05/19 17:39:30, reed1 wrote:
> 1. I suggest we keep the old version as well (even w/ the same name)
> 2. Lets try to define an object that holds all those arrays/counts/etc.

Sounds good to me.  Here's v2.

[reed1, 2016-05-19 18:21:00]

reed@google.com changed reviewers:
+ fmalita@chromium.org

[reed1, 2016-05-19 18:21:01]

seems like we can impl the older api in terms of this one trivially using stack
memory (which is nice).

How hard to try plumbing this up to actually work?

https://codereview.chromium.org/1992283002/diff/20001/include/core/SkCanvas.h
File include/core/SkCanvas.h (right):

https://codereview.chromium.org/1992283002/diff/20001/include/core/SkCanvas.h...
include/core/SkCanvas.h:938: uint32_t* fXDivs;
I would make all of these int instead of uint32_t, mostly just to make the
general skia style.

[msarett, 2016-05-19 19:18:47]

On 2016/05/19 18:21:01, reed1 wrote:
> seems like we can impl the older api in terms of this one trivially using
stack
> memory (which is nice).
> 
> How hard to try plumbing this up to actually work?
> 

Don't think it'll be too bad.  I'll give it a shot.

> https://codereview.chromium.org/1992283002/diff/20001/include/core/SkCanvas.h
> File include/core/SkCanvas.h (right):
> 
>
https://codereview.chromium.org/1992283002/diff/20001/include/core/SkCanvas.h...
> include/core/SkCanvas.h:938: uint32_t* fXDivs;
> I would make all of these int instead of uint32_t, mostly just to make the
> general skia style.

sgtm

[f(malita), 2016-05-19 19:24:20]

https://codereview.chromium.org/1992283002/diff/20001/include/core/SkCanvas.h
File include/core/SkCanvas.h (right):

https://codereview.chromium.org/1992283002/diff/20001/include/core/SkCanvas.h...
include/core/SkCanvas.h:956: *  (in the x-dimension) to fit into dst or must be
left "fixed".  The first patch
The last patch is also fixed, right?

[msarett, 2016-05-19 19:30:58]

https://codereview.chromium.org/1992283002/diff/20001/include/core/SkCanvas.h
File include/core/SkCanvas.h (right):

https://codereview.chromium.org/1992283002/diff/20001/include/core/SkCanvas.h...
include/core/SkCanvas.h:956: *  (in the x-dimension) to fit into dst or must be
left "fixed".  The first patch
On 2016/05/19 19:24:20, f(malita) wrote:
> The last patch is also fixed, right?

In the Android implementation, it really depends on the number of patches, since
they just alternate between "fixed" and "scalable".

In practical use cases, I think it is almost always:
First and last patch are both fixed
OR
First non-empty patch is scalable and last patch is also scalable

[f(malita), 2016-05-19 19:36:11]

https://codereview.chromium.org/1992283002/diff/20001/include/core/SkCanvas.h
File include/core/SkCanvas.h (right):

https://codereview.chromium.org/1992283002/diff/20001/include/core/SkCanvas.h...
include/core/SkCanvas.h:956: *  (in the x-dimension) to fit into dst or must be
left "fixed".  The first patch
On 2016/05/19 19:30:58, msarett wrote:
> On 2016/05/19 19:24:20, f(malita) wrote:
> > The last patch is also fixed, right?
> 
> In the Android implementation, it really depends on the number of patches,
since
> they just alternate between "fixed" and "scalable".
> 

Ah, that makes sense (I didn't parse the "alternating" bit off the bat).

[msarett, 2016-05-25 19:07:11]

Patchset #3 (id:40001) has been deleted

[msarett, 2016-05-25 20:15:15]

Description was changed from

==========
Draft API for drawBitmapNine()

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
Add more flexible drawBitmapNine() API

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[msarett, 2016-05-25 20:15:25]

Patchset #3 (id:60001) has been deleted

[msarett, 2016-05-25 20:35:40]

PTAL.  I've added a software implementation, a test, and a
benchmark.  GPU implementation is coming next.

I kept the codepath separate from the original
drawBitmapNine() because the construction of
SkNinePatchIter is a lot more involved for the new API.  We
have the option to combine the codepaths and see how that
looks as well.

8888 Performance Results (z620):
New Impl    Android Impl (with logging turned off)
440 us      440 us

GPU Performance Results (z620):
New Impl    Android Impl (with logging turned off)
27.4 us     36 us

I'm not really sure how to explain the GPU speed-up.  I
haven't written any GPU specific code yet.  Both
implementations call drawBitmapRect().

[reed1, 2016-05-31 15:29:57]

Seems like a good start.

Will be interesting to see how you override this for pictures (SkRecord). How
large (typically) are the run-counts? If they're large, I wonder if it would
ever help to store everything in SkData in some known format:

e.g. x-count, y-count, xruns[], yruns[]

That way the caller, having built this SkData, will just get it ref'd during
picture record, rather than memcpy'd... Just a thought.

[msarett, 2016-07-13 17:34:50]

Patchset #5 (id:120001) has been deleted

[msarett, 2016-07-15 14:20:57]

On 2016/05/31 15:29:57, reed1 wrote:
> Seems like a good start.
> 
> Will be interesting to see how you override this for pictures (SkRecord). How
> large (typically) are the run-counts? If they're large, I wonder if it would
> ever help to store everything in SkData in some known format:
> 
> e.g. x-count, y-count, xruns[], yruns[]
> 
> That way the caller, having built this SkData, will just get it ref'd during
> picture record, rather than memcpy'd... Just a thought.

Added implementations for SkRecorder and SkPictureRecord.  Let me know what you
think.  Any other SkCanvas subclasses that I need to override?

[reed1, 2016-07-15 16:27:04]

https://codereview.chromium.org/1992283002/diff/160001/include/core/SkCanvas.h
File include/core/SkCanvas.h (right):

https://codereview.chromium.org/1992283002/diff/160001/include/core/SkCanvas....
include/core/SkCanvas.h:974: void drawBitmapNine(const SkBitmap& bitmap, const
NinePatchDivs& divs, const SkRect& dst,
"Nine" is sort of a misnomer here

[msarett, 2016-07-15 16:38:34]

What do you think about drawNPatch() and NPatchDivs?

[reed1, 2016-08-01 16:26:02]

lgtm w/ name suggestions

https://codereview.chromium.org/1992283002/diff/160001/include/core/SkCanvas.h
File include/core/SkCanvas.h (right):

https://codereview.chromium.org/1992283002/diff/160001/include/core/SkCanvas....
include/core/SkCanvas.h:941: struct NinePatchDivs {
drawBitmapLattice

https://codereview.chromium.org/1992283002/diff/160001/include/core/SkCanvas....
include/core/SkCanvas.h:945: int* fXDivs;
const

[msarett, 2016-08-01 21:18:11]

Description was changed from

==========
Add more flexible drawBitmapNine() API

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
Add drawImageLattice() and drawBitmapLattice() APIs

The specified image/bitmap is divided into rects, which
can be draw stretched, shrunk, or at a fixed size.  Will be
used by Android to draw 9patch (which are acutally N-patch)
images.

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

[msarett, 2016-08-01 23:06:37]

The CQ bit was checked by msarett@google.com to run a CQ dry run

[commit-bot: I haz the power, 2016-08-01 23:06:45]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-08-01 23:08:15]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-08-01 23:08:16]

Dry run: Try jobs failed on following builders:
  Build-Ubuntu-GCC-Mips-Debug-Android-Trybot on master.client.skia.compile
(JOB_FAILED,
http://build.chromium.org/p/client.skia.compile/builders/Build-Ubuntu-GCC-Mip...)
  Build-Ubuntu-GCC-x86_64-Release-Trybot on master.client.skia.compile
(JOB_FAILED,
http://build.chromium.org/p/client.skia.compile/builders/Build-Ubuntu-GCC-x86...)
  skia_presubmit-Trybot on master.client.skia.fyi (JOB_FAILED,
http://build.chromium.org/p/client.skia.fyi/builders/skia_presubmit-Trybot/bu...)

[msarett, 2016-08-01 23:18:27]

Patchset #10 (id:240001) has been deleted

[msarett, 2016-08-01 23:19:00]

The CQ bit was checked by msarett@google.com to run a CQ dry run

[commit-bot: I haz the power, 2016-08-01 23:19:05]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[msarett, 2016-08-01 23:25:10]

https://codereview.chromium.org/1992283002/diff/160001/include/core/SkCanvas.h
File include/core/SkCanvas.h (right):

https://codereview.chromium.org/1992283002/diff/160001/include/core/SkCanvas....
include/core/SkCanvas.h:941: struct NinePatchDivs {
On 2016/08/01 16:26:02, reed1 wrote:
> drawBitmapLattice

Done.

https://codereview.chromium.org/1992283002/diff/160001/include/core/SkCanvas....
include/core/SkCanvas.h:945: int* fXDivs;
On 2016/08/01 16:26:02, reed1 wrote:
> const

Done.

https://codereview.chromium.org/1992283002/diff/160001/include/core/SkCanvas....
include/core/SkCanvas.h:974: void drawBitmapNine(const SkBitmap& bitmap, const
NinePatchDivs& divs, const SkRect& dst,
On 2016/07/15 16:27:04, reed1 wrote:
> "Nine" is sort of a misnomer here

Acknowledged.

[commit-bot: I haz the power, 2016-08-01 23:31:21]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-08-01 23:31:22]

Dry run: Try jobs failed on following builders:
  Test-Ubuntu-GCC-GCE-CPU-AVX2-x86_64-Debug-Trybot on master.client.skia
(JOB_FAILED,
http://build.chromium.org/p/client.skia/builders/Test-Ubuntu-GCC-GCE-CPU-AVX2...)

[msarett, 2016-08-02 13:54:31]

msarett@google.com changed reviewers:
+ halcanary@google.com

[msarett, 2016-08-02 13:54:32]

Needs an override for SkPDFCanvas.  Adding Hal.

[hal.canary, 2016-08-02 13:56:19]

PDF: 1gtm

[msarett, 2016-08-02 14:40:12]

Patchset #12 (id:300001) has been deleted

[msarett, 2016-08-02 14:40:47]

The CQ bit was checked by msarett@google.com to run a CQ dry run

[commit-bot: I haz the power, 2016-08-02 14:40:53]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[msarett, 2016-08-02 14:41:15]

Rewrote impls for picture serialization

[commit-bot: I haz the power, 2016-08-02 15:00:03]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-08-02 15:00:04]

Dry run: This issue passed the CQ dry run.

[msarett, 2016-08-02 15:04:48]

The CQ bit was checked by msarett@google.com

[msarett, 2016-08-02 15:04:48]

The patchset sent to the CQ was uploaded after l-g-t-m from reed@google.com
Link to the patchset: https://codereview.chromium.org/1992283002/#ps320001
(title: "Rewrite picture impls")

[commit-bot: I haz the power, 2016-08-02 15:05:00]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-08-02 15:05:59]

Description was changed from

==========
Add drawImageLattice() and drawBitmapLattice() APIs

The specified image/bitmap is divided into rects, which
can be draw stretched, shrunk, or at a fixed size.  Will be
used by Android to draw 9patch (which are acutally N-patch)
images.

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...
==========

to

==========
Add drawImageLattice() and drawBitmapLattice() APIs

The specified image/bitmap is divided into rects, which
can be draw stretched, shrunk, or at a fixed size.  Will be
used by Android to draw 9patch (which are acutally N-patch)
images.

BUG=skia:
GOLD_TRYBOT_URL=
https://gold.skia.org/search2?unt=true&query=source_type%3Dgm&master=false&is...

Committed:
https://skia.googlesource.com/skia/+/c573a40ed5024b463e47088d307e3164a486dba5
==========

[commit-bot: I haz the power, 2016-08-02 15:06:00]

Committed patchset #12 (id:320001) as
https://skia.googlesource.com/skia/+/c573a40ed5024b463e47088d307e3164a486dba5